reground/background colours?

Place your Xdefaults in $HOME/Mosaic or /usr/X/lib/app-defaults/Mosaic,
or $HOME/.Xdefaults-nodename. The desktop seems to control .Xdefaults.

Mosaic*homeDocument: http://myhome.com
Mosaic*Background: color-name-here
Mosaic*Foreground:  #000000000000

You can also set various Mosaic defaults on the command line, as in:

Mosaic -mono &   # for monochrome monitors

Mosaic -bg Green -fg White &  #  set the background colour to green and
     # the foreground text to white

        T15) Where can I find  a driver for my SVGA card ?

If you are running a release before 2.03, then
Checkout the betaX distribution on
ftp.freebird.org:/unixware/freebird/x11/servers/betaX.
Note that betaX is no longer updated.

        T16) How do I change from a 2 button to 3 button mouse?

Choose the Preferences/Mouse icon from the desktop
and select the appropriate number of buttons.

If you have a Logitech mouse, then you may also need to apply
tf2071.tar from ftp.sco.com

        T17) Howto update the default modem speed from 38.4 to 115.2

Bob Stewart has written a piece of code called hispeed.c included
here.  The source is attached (with permission).

This program changes the asycspdtab table in the running kernel
to support higher speeds than 38400 baud.  It changes the speed
of the B50 entry to 112500, and the B75 entry 50 57600.

There is a restrictions:

It only works on a 16550 UART PC serial port, ie: COM1-4

A binary version in uuencode format is available from
ftp.freebird.org:/unixware/freebird/hints/Uucp/hispeed.uu

===========================================================================

/*
 * SCCSID @(#)hispeed.c 1.5 9/27/95
 *
 * file: hispeed.c
 *
 * Copyright 1995, Bob Stewart
 * This program may be used freely with no limitations on distribution,
 * except that changes MUST be noted.
 *
 * You may not charge another person any fee for a copy of this program.
 *
 * Bob Stewart will not be responsible for any errors or problems caused
 * by this program whether through its proper use, misuse, or abuse.
 * It is your responsibility to examine this program, and decide whether
 * you wish to compile and run it on your system.  By doing so, you
 * accept full responsibility for any damage that may occur to your system.
 *
 * This program is not not warranteed to perform any function, including
 * the service it was designed to perform.  :-)
 *
 * NOTICE: This program modifies your running kernel!
 *
 * On Unixware 1.1.x systems, you MUST be using the asyhp driver.
 * On Unixware 2.x systems, you MUST be using the asyc driver.
 *
 * You MUST be using a 16550 uart for hispeed to be effective.  If you
 * are not using a 16550, you will probably experience data over-runs
 * and data loss at 115.2, and probably at 57.6, as well.
 *
 * This program changes the asycspdtab table in the running kernel
 * to support higher speeds than 38400 baud.  It changes the speed
 * of the B50 entry to 115,200, and the B75 entry to 57,600.
 *
 * These changes are NOT permanent.  The old speeds will return on reboot.
 * This program MUST be run by root.
 *
 * Change your /etc/uucp/Systems file to enable the higher speed, e.g.:
 *
 * dial-earth Any ACU 50 1235678 "" \d "" \d in:--in: login word: password
 * dial-sun   Any ACU 75 3193856 "" \d "" \d in:--in: login word: password
 *
 * Make this program by typing "make hispeed", then run it by typing
 * "./hispeed".  For 1.1.x systems, use "./hispeed -h".
 *
 * For a printout of the before and after table values, use "./hispeed -v",
 * or ./hispeed -v -h".
 */

#include
#include
#include
#include
#include

void main(int argc, char *argv[])
{
   int i;
   int verbose=0;
   char fileid[]="/dev/kmem";
   char symname[15];
   short speed[16];
   int  kmemfd;
   struct mioc_rksym rks;

   strcpy(symname,"asycspdtab");

   for(i=1; i<argc; i++)
   {
      if(argv[i][0] != '-')
         continue;

      switch(argv[i][1])
      {
         case('v'):
            verbose=1;
            break;
         case('h'):
            strcpy(symname,"asyhpspdtab");
            break;
         default:
            break;
      }
   }

   if((kmemfd = open(fileid, O_RDWR)) == -1)
   {
      perror("open /dev/kmem:");
      exit(-1);
   }

   rks.mirk_symname = symname;
   rks.mirk_buf = &speed;
   rks.mirk_buflen = sizeof(speed);

   if(ioctl(kmemfd, MIOC_READKSYM, &rks) == -1)
   {
      perror("ioctl /dev/kmem:");
      exit(-1);
   }

   if(verbose)
   {
      printf("Original %s\n",symname);
      for(i=0; i<16; i++)
      {
         printf("%4.4X ",speed[i]);
         if(i%4 == 3)
            printf("\n");
      }
      printf("\n");
   }

   speed[1] = 1;                /* change B50 to 115,200 baud */
   speed[2] = 2;                /* change B75 to  57,600 baud */

   if(verbose)
   {
      printf("Updated %s\n",symname);
      for(i=0; i<16; i++)
      {
         printf("%4.4X ",speed[i]);
         if(i%4 == 3)
            printf("\n");
      }
      printf("\n");
   }

   if(ioctl(kmemfd, MIOC_WRITEKSYM, &rks) == -1)
   {
      perror("ioctl /dev/kmem:");
      exit(-1);
   }

   close(kmemfd);

   fprintf(stderr, "hispeed installed\n");
}

        T18) How to setup  UUCP over TCP/IP between two UnixWare machines

Step 1. Edit /etc/uucp/Devices and uncomment the following line

CS  - - - CS

This then uses the connection server to make connections.

Adding ",eg" to the CS string, you can use the "e" =error-free protocol instead
of the much slower traditional "g" one.  As the TCP "line" can be guaranteed
as error-free, there is no need to use the "g" protocol if both sides support
the "e" protocol:

CS,eg  - - - CS

Step 2. Edit /etc/uucp/Systems , add a line for each system you want to talk
to, for example to system raven

raven     Any CS - -,listen:10103

This entry makes "uucico" run over the TCP network.  It causes the Connection
Server to connect to the "10103" service through the "listen" service.
The "10103" service invokes uucico.

Step 3. Edit /etc/uucp/Permissions, to allow remote systems
connecting over the connection server to send and request files (note
that remote systems also need an entry in /etc/uucp/Systems).

LOGNAME=nuucp SENDFILES=yes REQUEST=yes

Step 4.  When the above steps are completed on both UnixWare machines, use
the following command to make sure that you can UUCP to the remote system.
You should get "Conversation Complete: Status SUCCEEDED" at the end of the
uutry output.  If not, something is not configured properly, or the remote
machine is not a UnixWare machine and you should use the other
configuration method outlined below to establish UUCP communication.

/usr/lib/uucp/uutry -r  remote-sys-name

        T19) How to setup  UUCP over TCP/IP between UnixWare  and  another O/S running UUCP.

By conventions most machines support Uucp over TCP on port 540.

Step 1. Allow incoming connections on uucp port 540.

To perform these steps, you should be logged into the desktop as
the system owner and should have an xterm open where you have
used the 'su' command to become root.

1.1     Set up the listener to listen on port 540 and test it.

1.1.1   Edit the /etc/inet/inetd.conf file and add the following
line to the end of it (all fields are delimited by tabs):

uucp    stream  tcp     nowait  nuucp   /usr/lib/uucp/uucico    uucico

This sets up a listener through the inetd port monitor to listen
for uucp and run uucico as user nuucp.

1.1.2   Stop and start the inetd service by issuing the following
commands:

sacadm -k -p inetd

sacadm -s -p inetd

1.1.3   Test the new listener by telnetting to port 540 as
follows:
telnet localhost 540

You should get a prompt back that says, "Shere=FOO" where "FOO" is the
node name of your UnixWare machine.  You can just close this window,
open a new window and 'su' to root again.  Or, if you wait a few moments
the connection will just timeout and give you a '#' prompt back.

The 'Shere' line that is displayed is uucp on your machine saying,
"Here I am!  Send me some uucp commands."  If you don't get the "Shere"
then something is set up incorrectly.

Step 2.  Setup the outgoing uucp connection using the Connection Server.

In /etc/uucp/Devices put

CS  - - - CS

In /etc/uucp/Systems  put

node Any CS - node.domain.name,uucp in:--in: Logname word: passwd

Where node is the UUCP name node.domain.name is a full domain name that
TCP/IP can see uucp is the service in /etc/inet/services (=540)
Logname is the login name. Passwd is the password.

Plus usual Permissions entries.

Step 3. Test the connection.

Assuming that the remote system is ready to accept uucp communication
from this UnixWare machine on port 540, we should get "Conversation
Complete: Status SUCCEEDED" at the end of a uutry.  Now uutry the system
by issuing the following command:

/usr/lib/uucp/uutry -r   name

Where  "name" is the name of the remote system as defined in the
/etc/uucp/Systems file.

If it doesn't work, something isn't configured correctly.  Make sure
that the remote system is ready to receive uucp communication.  This
often requires an entry in the remote system's /etc/uucp/Systems or
/etc/uucp/Systems.tcp file for your local system.  (In other words, put
the UnixWare's system name in the remote machines Systems file(s).)

        T20) How do I get my serial communications working?

To obtain a document on how to trouble shoot serial communications,
see  http://www.freebird.org/hints/comms/serial-comms.

        T21) In UW1.x in order to use the 16550 UART chips in my serial card,
    I needed to use the asyhp driver. Is this still necessary in UW2?

In UnixWare 2 you should NOT use the 'asyhp' driver.  All functionality
of the 1.x asyhp driver has been rolled into the new 'asyc' driver so
there is no need to modify the default serial driver in UnixWare 2.
This is discussed in the man pages for both asyc and asyhp.

        T22) How can I debug outgoing modem connections?

Outgoing calls are managed by the connection server [cs(1M)].  Both
cu(1), uucp(1), and ppp(1) use the connection server to dial.

When you click on either 'dial' from the GUI device setup window or
double click on a system icon, you are running 'cu' and can use this
technique to debug your problem.

Many every day problems (such as the remote line being busy) can
be debugged using the connection server debug mode.

To activate cs debug mode, enter the following at the command
line as root:

   # ps -eaf | grep cs
   # kill -9 (the process id of the /usr/sbin/cs process)
   # /usr/sbin/cs -d

This will start a log in /var/adm/log/cs.debug.

A good debugging technique is to have the 'cs -d' running and in a
separate terminal window, enter 'tail -f /var/adm/log/cs.debug'.  Then,
in a separate window do your cu or uucp request.

You'll see what is being sent (e.g. atdt5551212) and what is being
expected (e.g.  CONNECT) as well as some technical details about which
line is being selected and so forth.

At some point you may want to turn cs debug mode off to save space.

To do this, kill the currently running cs process as shown above
and restart cs using '/usr/sbin/cs' without the -d option.
Alternatively, you could continue to run cs in debug mode and
write a cron(1M) daemon that keeps the size of the
/var/adm/log/cs.debug file reasonable.

        T23) How can I quickly setup outgoing uucp?

A tar file containing key /etc/uucp configuration files can be found on
the freebird archive (ftp.freebird.org:
/unixware/freebird/hints/Mail/uucp.quick.tar.Z) . This will setup a
hayes compatible modem on COM1 for outgoing uucp only.  See the README
file included for details.

        T24) How can I access a Comms port from DOS running under UnixWare?

>From a shell prompt on a virtual console try:

                dos +acom2 (or +acom1)

or under X windows, from a shell prompt try:

                dos +acom2 +x (or +acom1)

from you desktop, in the applications folder, click mouse button 3 on
the DOS icon, and then select Options, then select a COM port.  From
there you can invoke this options for a single session by choosing
Start, or you can have the current options tied to that icon by choosing
Apply.

All of the above will fail if another program (like the port manager,
connection manager, uucp or cu) has the port locked.

        T25) Where do I find updates and patches for UnixWare 2?

These can be located from ftp.sco.com:/UW20

(The following is taken from the SCO announcment posting with their
permission):

CONNECTION INFORMATION  for updates and patches for UnixWare2

For anonymous ftp connection:

Directory Name: /UW20

ftp to ftp.sco.com
Login name: ftp
Password:  your email address

Nameservice Note:

The wu-ftp software used on ftp.sco.com requires that two kinds of
DNS resource records for your site have been propagated to SCO's
nameserver:

A Name-to-address mapping  -and-   PTR Address-to-name mapping

The lack of a propagated PTR record for your site is the most common
cause of problems with anonymous ftp to ftp.sco.com. PTR records
are also known as "pointer" records or "reverse" records.

For anonymous UUCP connection:

Directory name:  /usr/spool/uucppublic/UW20

For USA, Canada, Pacific Rim, Asia and Latin America customers:

Machine name:  sosco
Login name:  uusls  (fourth character is the letter "l")
No password

List of modems available for UUCP transfer from sosco:

V32, V32bis                      5@   +408 425-3502
Telebit Trailblazer                   +408 429-1786

For Europe/Middle East/Africa customers there is a system located at
SCO EMEA (London):

Machine name:  scolon
Login name:  uusls
Password:  bbsuucp

List of modems available for UUCP transfer from scolon.sco.com:

Dowty Trailblazer  +44 (0) 1923 210911
V32                +44 (0) 1923 222681

For SCO Online Support (SOS) BBS download:

NOTE:  Access to the UnixWare Supplements will be available in the
       near future on the SCO Online Support (SOS) System.

These supplements can be downloaded interactively via XMODEM, YMODEM,
ZMODEM or Kermit.  Follow the menu selections under "Toolchest" from
the main SOS menu.

List of modems available for interactive transfer from SOS:

V32, V32bis  8@                 +408 426-9495

Telebit Trailblazer             +408 426-9525

Note:  telnet access to SOS is available by telneting to sos.sco.com

For customers with CompuServe Access:

Type "GO UNIXWARE"

For ftp via World Wide Web:

URL to open:  ftp://www.sco.com


        T26) How can I set the machine to auto reboot on a panic ?

Change /etc/default/init from

PANICBOOT=NO

to

PANICBOOT=YES

and remove the execute bit from /sbin/dumpcheck

        chmod -x /sbin/dumpcheck

Rebuild the kernel and reboot.
        /etc/conf/bin/idbuild -B
        /etc/shutdown -i6 -g60 -y

        T27) Does UnixWare-x.x support Card and Socket Services ?

   No.

        T27.1) Does UnixWare support PCMCIA?

   Yes and No. Anything that is supported or is known to be working only works
   with notebooks with Intel-compatible PCMCIA controllers (ie. Intel, Cirrus,
   Vadem, VSLI, etc...). See 'What PCMCIA cards work with UnixWare'

   Check with your notebook's vendor before making a purchase, ie.
   1/ What PCMCIA controller chip reside on the notebook ?
   2/ Is the PCMCIA controller chip Intel-compatible ?
   Note that some sales(wo)men just don't know what you are talking about.
   Double make sure that you can return the notebook.

   (I am willing to test new enablers/drivers/cards. Send your comments and/or
    corrections to: Thanh Ma at tma@encore.com)

        T27.2) What PCMCIA cards work with UnixWare?

   1/ UnixWare-1.x, UnixWare-2.0x:

      o Third party support:

        - Ositech Communication (http://www.ositech.com)
          Ethernet/Modem/Sound

        - Accton Technology - EN2212/EN2216 (http://www.accton.com)
          Ethernet only
          Note: the enabler was written by Lynsoft. It uses native UnixWare
                ne2k/NE2000 driver. This enabler is not supported by the vendor
                although it worked fine for me.

        - Socket Communication - Socket EA+ (http://www.socketcom.com)
          Ethernet and Serial
          Note: The enabler uses native UnixWare ne2k/NE2000 driver.
                This enabler is not supported by the vendor although it worked
                fine for me.

      o SCO support: None

   2/ UnixWare-2.1[x]:

      o Third party support:

        - Ositech Communication (http://www.ositech.com)
          Ethernet/Modem/Sound

        - Accton Technology - EN2212/EN2216 (http://www.accton.com)
          Note: see below

        - Socket Communication - Socket EA+ (http://www.socketcom.com)
          Note: Both of these cards are listed in the UnixWare-2.1 Hardware
                Compatibility Handbook (on the web site) as 'Reported' to work
                However, I failed to get it to work. The transmogrified
                NE2000 driver in UnixWare-2.1 has changed in such a way that
                none of these enablers could talk to the driver.

      o SCO support (http://www.sco.com)

        - Ethernet: 3Com-3C589x, Xircom-

        - Modem: most of the modems

        (Tested modem cards:US Robotics, Hayes, Megahertz)

Note: PCMCIA cards must be resident at boot time in order for them to work.


        T27.3) How do I get my XXXX modem card working ?

   Read 'Chapter E', section 'Configuring PC Card (PCMCIA) Modems'
   in your 'SCO Unixware - Installation Handbook'

        T27.4) How do I get my 3Com 3c589x or the Xircom ethernet card working ?

   Just like when you install your ISA ethernet card...

   1/ 3Com 3c589x: insert the card into the socket before installing
      UnixWare-2.1. Select the appropriate driver when prompted.  Or
      run 'niccfg' (available when you install the 'nics' package),
      and select the appropriate driver when prompted.

   2/ Not sure about the Xircom. Should be similar but I just never tried.

        T27.5)  Are there any pitfalls when installing UnixWare  2.x on a
   laptop with PCMCIA support?

(Martin Sohnius writes)

Yes, and yes.  :-)

Buy an Ositech Trumpcard "Jack of Diamonds" combined modem/ethernet card.
It's not cheap, but know to be the only one which works with UnixWare.  It
comes with a UnixWare driver in the box (which installs like a dream), and
according to the very good instructions in the README files, you can even
use it for a network install.

Don't install the nics package; the Ositech driver takes its place later.

You other question:  pitfalls.  It depends on the Thinkpad model.  Some of
them (notably the 755CX which I have, but also Pentium-upgraded other
models) fall over with the Machine Check Exception panic after the first
boot floppy.  Download the tf2069.tar file from the ftp server (*not* the
tf2069as.tar file -- that seems to be corrupt!) and create a new boot floppy
from it.  Follow the instructions in the *.txt file as to the special boot
parameter

        CR4_MCE=NO

You cannot use a CD-ROM in the Thinkpad proper (it's an IDE-ATAPI).  So,
unless you want to network install with your Ositech or with the "slip
install" in the USLE archive (takes forever!), you need a docking station
with a SCSI CD-ROM drive.  The SCSI adapter in "Dock 1" is a Future Domain
(driver name fdeb) and is factory set to IRQ 11.  Since UnixWare's default
is IRQ 5 for the fdeb driver, you must change it in the DCU (or set the IRQ
to 5 in the hardware).

Please, totally ignore the TID in the NetWire archive on 755E and docking
stations!  It's garbage.

        T27.6)  What laptop video chipsets does UnixWare support?

UnixWare has native drivers for:

1/ WD90C24xx
2/ Cirrus 54xx

If you have the GD7431/3 then you  may want to try 2/ above.
If the above does not work for you then try:

3/ XFree86 (http://www.xfree86.org)
4/ XInside (http://www.xinside.com)
5/ MetroLink (http://www.metrolink.com)

        T28) Are there any problem in getting UnixWare/NetWare connections
up and running?

Greg Smith (gsmith@westnet.com) writes:

One thing that I (and a number of other people) had problems with is
getting the UnixWare/NetWare connections up and running.  Specifically,
the NUC NLM module doesn't load correctly with older versions of NetWare
until a number of updates are taken care of.

I had so many problems with it that, after I solved all of them, I wrote a
guide on troubleshooting the initial problems for my own use before I
forgot any of it, and I packaged it up for others to use as well.  It's
about 5 pages long and available on my home page,
URL http://www.westnet.com/~gsmith (This is also available
on ftp.freebird.org:/unixware/freebird/hints/UW2/uw2nw).

        T29)  When installing the 2.02 update, an error appearead about
a bad entry on the contents file - how can this be fixed?

When trying to install the 2.02 update on to my 2.01AS system,
partway through the following error appeared:

 bad entry read from contents file :     - problem: unknown ftype
 UX:pkginstall: ERROR: unable to merge package and system information

The install halted after the error.  I've also noticed that on system
startup, the console shows an error message somewhat similar, about a
problem merging contents or something.  What's the cause of the
problem?

The problem is in the file /var/sadm/install/contents.  Usually the very first
line is badly corrupt (no idea what combo of circumstances causes this,
though).  Delete that first line.

        T30) How do I install the *.pkg.tar.Z files from the freebird archive?

        xxxx.pkg.tar.Z     A compressed tar archive of a binary in
                           pkgadd format

Where a file is suffixed with .gz this means that the GNU Zip
file compression utility has been used - to uncompress in this
case use gunzip.

The ".pkg" software package files follow the ABI (System V Application
Binary Interface) conventions for add-on
packages and are installed in /opt. It is wise to
change the default PATH setting to include /opt/bin in /etc/profile.

        PATH=$PATH:/opt/bin
        export PATH

If you do not have a /opt, you can create it with the subdirectories
/opt/bin, /opt/lib, /opt/man -- you could also use a symbolic link
to some other part of disk if you like (for example /usr/local) which
can be useful if you get short of free disk space.

To make the manual pages accessible to the /usr/ucb/man command
set the MANPATH environment variable to include /opt/man,
a typical MANPATH might be:

        MANPATH=/usr/share/man:/opt/man
        export MANPATH

The packages (pkg.tar files) should be installed as follows:

1. Extract the tar archive into /tmp.

$ cd /tmp

If compressed (.Z) then

$ zcat package.tar|tar xvf -
$ su
# pkgadd -d `pwd`

Many of the utilities include desktop icons for point and click
operation:

2. Installing graphically

On UnixWare 1.x:
----------------
To install graphically on UnixWare (after you have installed the package)
        Select the System_Setup icon from the Desktop.
        Select Application_Setup from System_Setup
        Wait while the system catalogs the applications
        Select the package icon from Application_Setup to get to
            Application_Setup:package_name
       Select the package icon and Finally click on the Install_to_desktop
        Selecting Applications from the Desktop you should
        then see the package icon.

On UnixWare 2.x:

For UnixWare 2.0 and later, use the App_Installer utility in Admin_Tools
after doing the pkgadd.

        Select the Admin_Tools icon from the Desktop.
        Select the App_Installer icon from Admin_Tools
        Select the icon of the package from the "All Applications
                currently installed" window
        Select Show_Contents
        Select the installable icon and hit Copy_to_folder
        Select Apply, followed by Ok.

        T31) How do I configure support for WWW virtual hosts, so that
one machine can have several IP addresses ?

This answer has now moved to the Howto Setup Networking FAQ.

        T32) How do I setup the NFS automounter?

On the Client side, Create the file /etc/auto.direct

/var/mail      -hard,grpid,intr,rsize=1024,wsize=1024  windsor.ukb.novell.com:/var/mail
/rtmp   -hard,grpid,intr,nosuid,rsize=1024,wsize=1024   windsor.ukb.novell.com:/rtmp
# This example takes the resource from two servers so if one goes down
# the other will kick in.
/opt    -ro,soft,grpid,intr,rsize=1024,wsize=1024       usle.ukb.novell.com:/fs1/destiny windsor.ukb.novell.com:/export2/destiny
/fs/marlow/home      -rw,hard,grpid,intr,nosuid,rsize=1024,wsize=1024    marlow:/home

Edit /etc/rc3.d/S22nfs and enable the automounter to use the /etc/auto.direct
map (this is towards the end of the start section)

        # Mount all NFS files listed in /etc/vfstab
        /sbin/mountall -F nfs &

        # These modified for local use
        if [ -x /usr/lib/nfs/automount ]
        then
          # only start the auto mounter if our main server is up
          /usr/sbin/ping windsor 1 | grep alive >/dev/null 2>&1
          if [ $? = 0 ]
          then
            /usr/lib/nfs/automount -tl 1800 /- /etc/auto.direct&
          else
            echo "Windsor is not up - not starting automounter"
          fi
        fi

        ;;
'stop')

On the server side, edit /etc/dfs/dfstab

For example:

share -F nfs /rtmp&
share -F nfs -oro /export2/destiny &
# this example restricts the sharing to just a couple of machines.
share -F nfs -o rw=london.ukb.novell.com:eaton.ukb.novell.com:carrera.ukb.novell.com /var/mail&

        T33) How do I change the IP address and name of a UnixWare box?

Randy Seuss and Martin Sohnius write:

        uname -S new_node_name
        /etc/inet/menu

Which then takes you thru the screens necessary for assigning
IP addresses, routers, etc.  I believe that the above is
enough, PROVIDED you either reboot the system or at least run the
command

        # sh  /etc/rc2.d/S65loopback

(For more info as to why, look at the S65loopback script yourself.)
and then

        # /etc/inet/rc.restart

        T34) SMC NIC install & TF2047 (UW2.01)

Kevin R. Bulgrien (102174.1561@compuserve.com) reports:

There is a patch TF2047 that corrects a problem with the SMC NIC driver.  The
error that it fixes can lead to an install session failure:

If you are installing UW 2.01 on a system that is already attached to a network
which is in use, you will undoubtedly experience a system crash during the
portion of the setup which detects and installs support for the NIC.

To proceed with the installation without risk of this failure, disconnect the
system from the network until after the install is complete.

To proceed with the installation after it has locked up during the network
setup phase, reboot after you have disconnected the system from the
network.  Installation of the network card will proceed normally.

I experienced this failure a couple of times, and arrived at the above solutions
by trial, error, and reading over the TF listings - and noticed the SMC driver
patch.

This information is implied by the documentation of the TF2047 patch in that
it indicates that the SMC driver can crash the system if it receives packets
too quickly during the driver initialization, but I have not seen it
associated directly with problems experienced during an install.

        T35)How can I read the size of installed memory in a UnixWare system?

At the command line as root, use the memsize(1M) or prtconf(1M)
commands.

In C you can use the sysi86(SI86MEM) call, for example:

#include
#include

main()
{
printf ("memsize = %ld \n", sysi86(SI86MEM));
}

        T36) I have UnixWare 2.03 and have problems reading multiple files
from floppy (the machine has 32MB of RAM).

Apply ptf2202.tar from ftp.sco.com:/UW20/


        T37)How do I found out the number of max processes a UnixWare machine
is set for?

        su
        /etc/conf/bin/idtune -g NPROC

idtune then prints out at most four values

        The current value of the tunable parameter
        The default value, from the Mtune entry
        The minimum valid value, from the Mtune entry
        The maximum valid value, from the Mtune entry

To increase the number of processes to say 1200

        /etc/conf/bin/idtune NPROC 1200

        T38) I've lost the root password, what do I do now?

If you can login as the "system owner", then you can use tfadmin(1m)
to change root's password.  This works on both UnixWare 1.x and 2.x:

    $ /sbin/tfadmin passwd root

or to delete the root password (remember to reset it afterwards)

    $ /sbin/tfadmin passwd -d root

Otherwise you need to use the emergency boot floppy (an EBF).


        T39) How to recover with an Emergency boot?

Peter Lord writes:

At some time or other, you will find yourself with a UnixWare machine
which either you can't log into or won't boot up. Their are many
possible causes for this but perhaps the most common include forgetting
passwords, configuring the kernel incorrectly and editing system files
by hand incorrectly.

One example (now laughable - but not at the time!) was when I wanted to
bring the system into single user mode for admin. I edited /etc/inittab
to switch the default run-level to 1. Later, I put the run-level back to
6 (not 3!). This left me with a machine which continually re-booted
itself!

Creating an emergency boot disk set is an essential admin task. On
UnixWare 2, create a disk set by :-

     emergency_disk diskette1

This will take a few minutes. Note that this disk set is unique for you
machine. Also, as it contains all your HBA drivers, make sure you create
it after your machine is stable and fully installed.

When booting with your disk set, a mini version of UnixWare is booted
from the floppies. You will be presented with a short menu allowing you
to repair disks, mount disks and enter a shell. In my example above, I
could enter the shell and edit /etc/inittab before rebooting.

As well as creating an emergency boot set, you can also create a backup
of your whole machine onto tape and restore this tape with the emergency
boot set. See the  emergency_rec program to create such a tape.

        T40) Where are the system messages and log files on UnixWare?

Peter Lord Writes:

As a UnixWare system administrator, you need to keep track of the
various messages that the system generates. But where are these
messages?

Console Messages

If you are using the graphical desktop, you may miss some messages
because they are written out to the console or your home virtual
terminal (VT).

You can switch from the graph VT to the home VT by pressing ALT-SYSRQ
together, then H. To get back to the graphical VT press ALT-SYSRQ F1.

Operating system messages

Occasionally, the kernel will want to print some message out to you.
This may be some information, warning or error messages. There are a few
ways to read these messages :-

1. /dev/osm devices. As root, you may access the /dev/osm* devices to
view these messages. The command "cat /dev/osm1" will print out the messages to
date and the command "tail -f /dev/osm1" will print out the last few
messages and will continue to print out new ones.

2. /var/adm/log/osmlog file. This file records the output of /dev/osm.
Old versions of this file (from previous boots) are kept in the same
directory. This file is usually readable by all.

3. Msg_Monitor. This is a graphical tool which monitors the operating
system messages from the desktop (found in the Applications folder). This is
configurable (so, for example, you can set it up to de-iconify on
receipt of a new message).

System messages

Application programs also want to alert you to possible problems.
Usually these will either be generated at the console or through the
/dev/osm interface.

However, it is possible to collect these messages by severity into a log
file. In fact, all the messages from all the machines on the network may
be collected together into one file. This system is syslog.

The syslog daemon (syslogd(1)) is started from /etc/inet/config. Make
sure you have a line thus in this file :-

/etc/inet/config:8:/usr/sbin/syslogd::y:/etc/syslog.conf::

And the configuration file /etc/syslog.conf should look something like this :-

*.err;kern.debug;daemon.notice;mail.crit;user.none      /var/adm/log/messages
*.alert;user.none                                       root
*.emerg;user.none                                       *
auth.info                                               /var/adm/log/authlog
*.debug                                                 @loghost

(Note in this example, loghost is an alias in /etc/hosts for the IS
monitoring machine)

Connection Server

The Connection server (cs(1)) is a daemon which establishes connections
for TLI/serial network services such as uucp, PPP etc. The log file for
the connection server is kept in /var/adm/log/cs.log.

However, if you are experiencing connection problems, the information in
this log file may not be detailed enough (for example, your uucp
connection is failing). The cs daemon can be run in debug mode to give
more information in the file  /var/adm/log/cs.debug by running cs with
the -d option.

Firstly, kill off the existing cs process (do a ps -edf | grep cs).
Secondly run cs -d (as root).

        T41) How can I revert to the BSD form of (restricted) chown?

By default, chown() system call comes with the old AT&T behavior and
allows a user to change the ownership of a file he owns to that of any
other user on the system.

How can I modify the behavior to the BSD-form (only root can change
the ownership of a file)?

The BSD way is the FIPS 151-2 and XPG4 way, and indeed there is a tuneable
called RSTCHOWN. For strict conformance (and when testing for
POSIX FIPS 151-2, XPG etc) this should be set to one.

/etc/conf/bin/idtune -g RSTCHOWN  will return its value.

To set it do

     # /etc/conf/bin/idtune RSTCHOWN 1
     # /etc/conf/bin/idbuild

and then reboot.

        T42) Why do I get 'syslog: accept: SIOCGPGRP failed errno 22' in syslog files ?

If you have setup syslog log you may sometime see messages like :

Aug  1 13:05:26 myhost syslog: accept: SIOCGPGRP failed errno 22

It's logged at level USER, thus you may not see it if you don't log
this level.

These messages are related to some program for which system libraries
were not specified in the correct order. A common sample is Sendmail
8.7.5.

If the program is compiled with C library before socket & nsl libs you
may get them.

To simply correct the problem, you have to find which program was
badly compiled and correct the Makefile. Just put -lc after -lnsl
-lsocket, and all should be fine.

        T43) How to turn off multiprocessors in a Unixware 2?

Try 'psradm':

      EXAMPLES
            To unlock a processor, or bring it online:

                        psradm -n processor_id

            To lock a processor, or take it off-line:

                        psradm -f processor_id

        T44) How to avoid the check for open virtual screens when loggin on the console?

Q Sometimes it happens that I can't login at the console
window ( h) while logged into the desktop.
I get a message that vt01 is still open and have the choice
to close it by hand. I then need to log off the desktop to
be able to get a login prompt on the console.

Does anybody know how to avoid the check for open virtual
screens?

A. Curt Smith writes:

tymon is a pain for too many reasons.  Here's my work around for
hacking both: /etc/inittab (this file is manufactured from /etc/conf/init.d/*)
and /etc/conf/init.d/consoles.  The later is a new file in that dir.

/etc/conf/init.d/consoles:

v1:23:off:/etc/getty vt01 console
v2:23:off:/etc/getty vt02 console
v3:23:respawn:/etc/getty vt03 console
v4:23:respawn:/etc/getty vt04 console
v5:23:respawn:/etc/getty vt05 console
v6:23:respawn:/etc/getty vt06 console
v7:23:respawn:/etc/getty vt07 console
v8:23:respawn:/etc/getty vt08 console

The above gives you virtual consoles on: Alt-SysReq-F[3-8]
Cat the above file to the end of /etc/inittab

Modify /etc/inittab and /etc/conf/init.d/kernel

Modify line to:
co:12345:off:/usr/lib/saf/ttymon -g -v -p "Console Login: " -d /dev/console -l console

Add line:
co1:12345:respawn:/etc/getty console console

BTW: for UnixWare 2.x and 2.1 I've found ttymon to be friendlier than UW 1.x and
not requiring being disabled as the above modification does.

You be the judge as to how far you want to free yourself from the
ttymon/sac/saf tyranny.

Alternately Anthony W. Southworth writes:

Personally, I don't have a problem with ttymon, but... the very first
thing I do after a new install is remove the '-v' option from the
console ttymon.  I don't see why it's the default configuration.
Without the '-v', you can login and logout of the home "console" without
shutting down the X server running on vt01.

The inittab file (and /etc/conf/init.d/kernel) should look something
like this:

co:12345:respawn:/usr/lib/saf/ttymon -g -p "Console Login: " -d /dev/console -l console

Here's the description of the '-v' option from the ttymon man page:

          Options
             In Express Mode (or standalone) ttymon takes the following
             options:

            -v     On startup, ttymon will list all open virtual terminals
                   (vts) associated with the device specified (typically
                   /dev/console).  On exit, the user will be prompted to
                   close all open vts either manually or automatically.
                   Choosing manual places the user in the highest numbered
                   vt at which point the user must manually exit from each
                   vt.  Choosing automatic will cause the signals SIGHUP
                   and SIGTERM to each open vt, then after 3 seconds
                   SIGTERM will be sent to each open vt to ensure
                   termination.

                   For systems running with the Enhanced Security Utilities
                   installed, the -v option will disable trusted path
                   processing.  [See ttyadm(1M)].

        T45) How to increase per-process size limits to avoid Virtual MEmory Exceeded errors?

Problem:
We have an application developed locally that needs as much memory
as possible to load 10-15MB digital map files. The default setup on
UnixWare 2.03 systems is 32 MB with 64 MB swap, but the program gets
a "Virtual Memory Exceeded" error around 14MB. Other tests show this
error occurs at 16MB predictably.

Solution. MS writes:

You need to up your process-size limits.  Either with the ulimit
or limit commands (for Bourne/Korn and C shell, respectively) on
a per-process base (only works for root), or by retuning the
kernel parameters

        # /etc/conf/bin/idtune SDATLIM 33554432         ## soft
        # /etc/conf/bin/idtune HDATLIM 33554432         ## hard
        # /etc/conf/bin/idtune SVMMSIZE unlimited       ## soft
        # /etc/conf/bin/idtune HVMMSIZE unlimited       ## hard
        # /etc/conf/bin/idbuild -B
        ...
        reboot!

The first two set the maximum size of the data segment (the "heap")
to 32 MB, the second two remove the limit on the total process size
(which is, however, implicitly limited by 32MB of heap + 16MB of
stack + program text).  Don't raise the stack limit, or a runaway
recursion will kill you.

----------------------------------------------------------------------------

Acknowledgements

Thanks go to the following for their contributions (either directly
or on the netnews) in putting this FAQ together:

        Leo Smith, Bob Stewart, Martin Sohnius, Greg Smith, Mike West,
Dan Busarow, Randy Seuss, Jeff Lind, Tim Rice, David Stone, Peter Lord,
Curt Smith, Anthony Southworth, Ian Hay,
Kevin R. Bulgrien, Thanh Ma for the PCMCIA section  and last but not
least, the Novell  & SCO staff who put the tech info on
the WWW.novell.com & ftp.sco.com sites.

----------------------------------------------------------------------------
Andrew Josey, To email remove #nospam from From: line.
Disclaimer: Any views expressed are not those of my employer, either past,
present or future.